DeepLab_V3 Image Semantic Segmentation Network

Implementation of the Semantic Segmentation DeepLab_V3 CNN as described at Rethinking Atrous Convolution for Semantic Image Segmentation.

For a complete documentation of this implementation, check out the blog post.

Dependencies

• Python 3.x
• Numpy
• Tensorflow 1.4.0

Downloads

Download the model checkpoints and dataset.

• checkpoints
• Dataset
  • Option 1
  • Option 2

Training and Eval

To train this model run:

python train.py --starting_learning_rate=0.00001 --batch_norm_decay=0.997 --gpu_id=0 --resnet_model=resnet_v2_50

Check out the train.py file for more input argument options. Each run produces a folder inside the tboard_logs directory (create it if not there).

To evaluate the model, run the test.py file passing to it the model_id parameter (the name of the folder created inside tboard_logs during training).

python test.py --model_id=16645

Datasets

To create the dataset, first make sure you have the Pascal VOC 2012 and the Semantic Boundaries Dataset and Benchmark datasets downloaded.

After, head to dataset/ and run the CreateTfRecord.ipynb notebook. The custom_train.txt file contains the name of the images selected for training. This file is designed to use the Pascal VOC 2012 set as a TESTING set. Therefore, it doesn't contain any images from the VOC 2012 val dataset. For more info, see the Training section of Deeplab Image Semantic Segmentation Network.

Obs. You can skip that part and direct download the datasets used in this experiment - See the Downloads section

Results

• Pixel accuracy: ~91%
• Mean Accuracy: ~82%
• Mean Intersection over Union (mIoU): ~74%
• Frequency weighed Intersection over Union: ~86.